Data Set htwddKogRob-InfDynSim for Localization in Highly Crowded Environments

This data set presents a major challenge for robot localization in highly crowded environments. The total distance travelled over all runs is 113.3 km. 50 dynamic obstacles (see htwddKogRob-InfDynSim_dynObstacles.png) were inserted into the map (see htwddKogRob-InfDynSim.png | 1px \(\widehat{=}\) 0.1m).

The work was first presented in:

A Fuzzy-based Adaptive Environment Model for Indoor Robot Localization

Authors: Frank Bahrmann, Sven Hellbach, Hans-Joachim Böhme

Date of Publication: 2016/10/6

Conference: Telehealth and Assistive Technology / 847: Intelligent Systems and Robotics

Publisher: ACTA Press

Additionally, we present a video with the proposed algorithm and an insight of this dataset under:

youtube.com/AugustDerSmarte

https://www.youtube.com/watch?v=26NBFN_XeQg

Instructions for use

The zip archives contain ascii files, which hold the log files of the robot observations and robot poses. Since this data set was recorded in a simulated environment, the logfiles include both a changed starting position and a ground-truth pose. For further information, please refer to the header of the logfile. To simplify the parsing of the files, you can use these two Java snippets:

Laser Range Measurements:

List ranges = new ArrayList<>(numOfLaserRays);
List errors = new ArrayList<>(numOfLaserRays);

String s = line.substring(4);
String delimiter = "()";
StringTokenizer tokenizer = new StringTokenizer(s, delimiter);

while(tokenizer.hasMoreElements()){
String[] arr = tokenizer.nextToken().split(";");
boolean usable = (arr[0].equals("0")?false:true);
double range = Double.parseDouble(arr[1]);

ranges.add(range);
errors.add(usable?Error.OKAY:Error.INVALID_MEASUREMENT);
}

Poses:

String poseString = line.split(":")[2];
String[] elements = poseString.substring(1, poseString.length()-1).split(";");
double x = Double.parseDouble(elements[0]);
double y = Double.parseDouble(elements[1]);
double phi = Double.parseDouble(elements[2]);